RadarCat presented at UIST2016

SACHI research project RadarCat (Radar Categorization for Input & Interaction), highlighted earlier this year in the University news, the Courier and Gizmodo and in a Google I/O ATAP 2016 session, will be presented at UIST2016 this week.

RadarCat is a small, versatile radar-based system for material and object classification which enables new forms of everyday proximate interaction with digital devices. SACHI’s contribution to Project Soli featured in a previous blog post SACHI contribute to Google’s Project Soli, in May. Read more about RadarCat for object recognition on the SACHI blog.

Google's Project Soli workshop in March 2016

Google’s Project Soli workshop in March 2016

Acacia – The Smart Image Compressor

Today we are releasing Acacia – a machine learning enabled image compressor developed here in the School of Computer Science at the University of St Andrews. Acacia is the work of Oleksandr Murashko and Dr. John Thomson.

screenshot

Acacia (Advanced Content-Adaptive Compressor of ImAges) is an image compression tool targeting at those who want the best compression under constrained energy or processing time scenarios – for instance a mobile device or a cloud image server. It allows users to target specific image quality or file size metrics when compressing an image with JPEG or WebP, with only minimal additional compression time. It does this by using machine learning to predict how an individual image will be compressed, and adjusts the aggressiveness of compression accordingly.

Acacia allows users to target compression to their file size or quality needs, significantly increasing the effectiveness of compression by adjusting to each individual image. It is available with a graphical interface, and with a CLI for batch processing.

Acacia is free and open source, runs on Windows, Linux and MacOS, and is available on Github as source, or as a Windows binary.

This software accompanies our paper, Predicting and Optimizing Image Compression, published in ACM Multimedia this week. The paper is available for free from John Thomson’s web site.

Visit by the new Principal and Vice-Chancellor ‌Professor Sally Mapstone

On Tuesday the 5th of October we were pleased to host our new Principal and Vice-Chancellor ‌Professor Sally Mapstone to visit the School of Computer Science. During this visit she was able to meet with staff and students, visit our teaching spaces and visit some of our research labs. We discussed our new Engineering Doctorate (EngD) in Computer Science, our PhD programme, our new and existing MSc programmes, our growth in undergraduate single, joint and MSci degree programmes along with changes to our teaching and research space over the past few years.

From left to right, Simon Dobson, Ruth Letham, Steve Linton, Sally Mapstone, Aaron Quigley, Robin Nabel and Dharini Balasubramaniam

From left to right, Simon Dobson, Ruth Letham, Steve Linton, Sally Mapstone, Aaron Quigley, Robin Nabel and Dharini Balasubramaniam

We were also able to showcase some of our ongoing research which included a short talk from Adam Barker, on Distributed Systems and his recent time with Google, and demonstrations from Chris Jefferson, on visualisation of constraints, Vinodh Rajan Sampath, on Scribal Behaviour and Digital Palaeography, Gonzalo Mendez, on iVolver, Gergely Flamich and Patrick Schrempf, on RadarCatHui-Shyong Yeo on WatchMi and David Morrison, on Beyond Medics.

We thank all the staff and students who made our new Principal feel welcome here in Computer Science.

 

 

Aaron Quigley appointed as ACM SIGCHI Vice President for Conferences

Congratulations to Professor Aaron Quigley who has been appointed to the ACM SIGCHI Executive Committee, to serve as the Vice President for Conferences. The ACM Special Interest Group on Human Computer Interaction (SIGCHI) is the premier international society for professionals, academics and students who are interested in human-technology & human-computer interaction. SIGCHI sponsors or co-sponsors 24 conferences in addition to providing in-cooperation support for over 40 other conferences. This family of HCI conferences are held across the year and around the world.

As Vice-President for conferences, Aaron will be responsible for strategic planning for SIGCHI-sponsored conferences, overseeing all aspects of SIGCHI-sponsored conferences, chairing various boards and committees and working with other SIGCHI vice-presidents and the SIGCHI executive committee on policies affecting SIGCHI sponsored, co-sponsored, and in-cooperation conferences.

img

PhD Viva Success: Michael Mauderer

Belated congratulations to Michael Mauderer, who successfully defended his thesis earlier this month. Micheal’s thesis, augmenting visual perception with gaze-contigent displays, was supervised by Dr Miguel Nacenta. Professor Aaron Quigley acted as internal examiner and Professor Hans Gellersen, from Lancaster University acted as external examiner.

mm

Multi-modal Indoor Positioning: Trends and Challenges by Prof. Niki Trigoni, Oxford University

Abstract:

GPS has enabled a number of location based services outdoors, but the problem of localisation remains open in GPS-denied environments, such as indoors and underground. In this talk, I will discuss the key challenges to accurate and robust position estimation, and will describe a variety of sensor modalities and algorithms developed at Oxford to address this problem.

The talk will cover inertial, radio-based and vision-based localisation approaches and their advantages and disadvantages in different settings.

 

Short Bio:

Niki Trigoni is a Professor at the Oxford University Department of Computer Science and a fellow of Kellogg College. She obtained her PhD at the University of Cambridge (2001), became a postdoctoral researcher at Cornell University (2002-2004), and a Lecturer at Birkbeck College (2004-2007). Since she moved to Oxford in 2007, she established the Sensor Networks Group, and has conducted research in communication, localization and in-network processing algorithms for sensor networks. Her recent and ongoing projects span a wide variety of sensor networks applications, including indoor/underground localization, wildlife sensing, road traffic monitoring, autonomous (aerial and ground) vehicles, and sensor networks for industrial processes.

Event details

  • When: 8th November 2016 14:00 - 15:00
  • Where: Cole 1.33
  • Series: School Seminar Series
  • Format: Seminar, Talk

Quicker Sort by Dietmar Kühl, Bloomberg L.P.

 

Abstract:

Quicksort is a well-known sorting algorithm used to implement sort functionality in many libraries. The presentation isn’t really about the algorithm itself but rather about how to actually create an efficient implementation of the algorithm: a text-book implementation of the algorithm actually is not that quick (even if the pivot is chosen cleverly). It takes paying some attention to detail to improve the implementation significantly. This presentation starts with a simple implementation and makes incremental improvements to eventually yield a proper generic and fast sorting function. All code will be in C++ but it should be possible to follow the majority of the reasoning with knowledge of another programming language.

 

Short Bio:

Dietmar Kühl is a senior software developer at Bloomberg L.P. working on the data distribution environment used both internally and by enterprise installations at clients. Before joining Bloomberg he has done mainly consulting for software projects in the finance area. He is a regular attendee of the ANSI/ISO C++ standards committee, presents at conferences, and he used to be a moderator of the newsgroup comp.lang.c++.moderated. He frequently answers questions on Stackoverflow.

Event details

  • When: 25th October 2016 14:30 - 15:30
  • Where: Cole 1.33
  • Series: School Seminar Series
  • Format: Seminar, Talk

Running Before We have Evolved Legs: The Gap Between Theory and Practice in Evolutionary Algorithms by Prof. John McCall

Abstract:

Evolutionary algorithms (EA) has developed as an academic discipline since the 1960s. The subject has spawned major subfields such as swarm intelligence and genetic programming and is applied to a wide variety of practical real world problems in science medicine and engineering. EAs are often the only practical method of solving large combinatorial optimisation problems and have achieved best-known results on a variety of benchmark problems. The global academic EA community is highly active, supporting several large international conferences and high-quality international journals. Despite this activity, sustained over decades, the community has struggled to make significant progress on developing a satisfactory theory of EAs. At the same time, substantial progress has been made on developing more sophisticated EAs that are ever more powerful but ever less amenable to theoretical study. In this talk I will outline some of the main approaches to a theory of EAs and illustrate the gap between those EAs that can be theoretically analysed by those approaches and EAs that are being used in practice. I will conclude with some interesting current developments and key open questions.

 

Short Bio:

John McCall is a Professor of Computing Science at Robert Gordon University.  He works in the Computational Intelligence research group, which he founded in 2003. He has over twenty years research experience in naturally-inspired computing.  His research focuses on the study and analysis of a range of naturally-inspired optimization algorithms (genetic algorithms, particle swarm optimisation, ant colony optimisation, estimation of distribution algorithms etc.) and their application to difficult learning and optimisation problems, particularly real-world problems arising in complex engineering and medical / biological systems. Application areas of this research include medical decision support, data modeling of drilling operations, analysis of biological sequences, staff rostering and scheduling, industrial process optimization and bio-control. He has over 90 publications in books, journals and conferences. He has successfully supervised 13 PhD students and has examined over 15 PhD theses.

Event details

  • When: 11th October 2016 14:00 - 15:00
  • Where: Cole 1.33
  • Series: School Seminar Series
  • Format: Seminar

David Harris-Birtill: Converge Challenge Winner 2016

Congratulations to Dr David Harris-Birtill, who was announced the winner of the Converge Challenge KickStart award at a ceremony in Edinburgh yesterday. The converge challenge competition rewards an early-stage idea or a new product. David won a cash injection prize of £3,000 to kick-start Beyond Medics – Automated Remote Pulse Oximetry, a camera based system that remotely measures patients’ vital signs.

david_hb_cc_winner2016

Deepview Project: Innovative GAZER Software

Congratulations to Dr Miguel Nacenta and Michael Mauderer on the success of Deepview and its subsequent application Gazer, an open source tool that provides functionality for showcasing light field images using gaze-contingent focus. The software, developed by SACHI, works in conjunction with eye tracking devices to allow photographers using light field cameras to discern images by automatically concentrating on objects using just their eyes.

Gazer-Logo-full-768x202

Results from the project have been widely disseminated in the media and featured on a BBC click episode (20.56)

For more information and downloads visit the Gazer project section or github repository.

The project was funded through the European Union’s Marie Curie Program (CIG – 303780).